Date of Award

Summer 2017

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

Mechanical Engineering

First Advisor

Allen, Casey

Second Advisor

Singer, Simcha L.

Third Advisor

Bowman, Anthony J.

Abstract

There have been discrepancies noted with regards to experimental data from rapid compression machines (RCM). When data is compared from different RCM facilities, the ignition delay times are inconsistent when inspecting any particular temperature. Currently in publications, if these datasets are compared, the discrepancy is said to be due to heat loss, however this issue has yet to be examined more thoroughly. To determine what the root cause of this discrepancy is, four different fake RCM facilities were created and simulated. There were also different sets of initial conditions used to determine how this may affect the data. They were simulated using a Multi-Zone Model, which is a one-dimensional model that uses a piston trajectory to calculate the change in volume over time to define the pressure in the reaction chamber for a given set of initial conditions. To assist in determining which initial conditions to use for any combination of desired compressed conditions, an Artifical Neural Network was used. There was a different network created for each machine, and they were trained to be able to predict the compressed temperature and pressure given a set of initial conditions. Once the initial conditions were determined, the simulations were run and the data was analyzed. It was determined that the compression time was the most important geometric factor leading to the discrepancy. It was also determined that the most influential set of initial conditions involved changing the initial pressure of the mixture as well as the compression ratio to reach the desired values.

COinS